Viewpoint representation in object recognition: Evidence from repetition blindness